This is the first study investigating preferred (self-selected) step frequency (PSF) in relation to optimal step frequency (OSF) during running at different speeds outside the laboratory. OSF was defined as the SF at which heartrate (HR) showed a minimum during running at a constant speed. We aimed for a high ecological validity, which prohibits the direct measurement of oxygen consumption. HR was used as a proxy of energy cost, as has previously been validated. After a practice session, 21 experienced recreational runners (12 female) ran in an out-and-back manner on a straight flat tarmac lane in three main sessions: (1) 27 min at fixed comfortable speed with PSF, (2) at the same speed but with different imposed SFs (range 140-200 steps min −1) and at PSF and (3) similarly but at 15% higher speed. Runners were paced by a cyclist. Second order polynomials fitted the individual SF-HR relations well. At comfortable speed (3.04 ± 0.28 m s −1) mean r 2 was 0.81 (range: 0.40-0.99) and PSF and OSF respectively were 165 ± 8 and 171 ± 8 steps min −1 (p < .05). Both increased (p < .05) by 3 steps min −1 at the 15% higher speed. SFs were negatively (−0.66 < r < −0.49, p < .05) related to body height. From the individual SF-HR relations we deduced that changing PSF to OSF would decrease median HR only by 0.5 beats min −1 , but for two participants meaningful reductions (1-3 beats min −1) were predicted. Outdoors, experienced recreational runners ran at a PSF slightly below OSF, but for most of them, there was little to gain in terms of HR reductions by increasing their SF to OSF.